282 ALASKAN GLACIER STUDIES Below this lake the stream course was partly marginal to the present ice sheet for a short distance, then came several sections in broad gorges through the drift and rock with some waterfalls, alternating with broader, lake-like sections where only the outer moraine separated the stream from the present ice edge. Here the lobate eastern margin of the glacier formerly forced the stream into marginal channels across rock spurs, making lakes above the spurs and rapids on the rocks, and these conditions still lingered, though the stream had cut somewhat in the rock. Below the westernmost gorge the stream emerged into a muddy tidal estuary which branched from the bay east of Heather Island in 1909, while across a part of it in 1910, the advance of the glacier had formed another lake with a till dam, as already described. The Main Glacier Surface. Columbia Glacier surface slopes about 150 feet to the mile. The glacier margins discussed in the preceding pages represent only the outermost portions, and do not include those in the upper part of the glacier at a distance back from the sea. These portions of the glacier, and the low grade, main glacier surface, we saw only from the crest of Heather Island and from the 1497 foot station southeast of the glacier. The conditions can be described fairly well, however, for the viewpoints were comprehensive arid we were able to write notes and take photographs from stations at which similar records had been made ten years before. The main glacier surface was impassably crevassed in each of the periods of observation, though it was more severely crevassed in 1909 and 1910 than ten years before, and on the east side there were one or two ice steps with especially severe crevassing. The changes in the medial and lateral moraines, and in crevassing within them, are the most important differences between the 1899 glacier surface and that which we saw, for the changes accompanying the thickening and spreading of the margins were not determin-able in the main glacier back of the frontal portion. Gilbert has described the conditions in 1899 as follows:1 "Opposite the great nunatak were two medial moraines, one passing within a hah* mile of its base, the other lying about one mile from the opposite edge of the glacier. A central tract two miles broad was practically drift-free. Toward the end of the glacier this central tract broadened, the medials swinging toward the sides, until finally the white belt was three miles wide. As the medials diverged they also broadened, and they eventually merged with1 flanking moraines, so that near the end, especially on the east side, the areas of drift-covered ice were very wide. The marginal belt on the west, instead of continuing northward parallel to the medial with which it was associated, was seen to curve about into the western embayment, as indicated on the map, and a belt seen from a distance near the north edge of the embayment was supposed to be its continuation, a loop being made within the embayment. As the ice in the embayment descended toward the west, it is evident that the morainic loop could not at that time represent a line of continuous flow, for we cannot suppose the ice to flow into the embayment on a descending course and then return on a parallel ascending course. It is therefore probable that the moraine was formed as a comparatively direct line, of drift, following the course of the main current at a time when no current entered the embayment. The inference that a change has occurred naturally leads to inquiry as to the precise nature of the antecedent condition of the glacier. On the one hand, the embayment may have been so full of ice that the surface gradient was outward; or, on the other, the glacier of the main valley ťOp. cit., pp. 74-5.